first version! help
trained on various past generations from midjourney; V1 is trained on roughly a hundred images, mini is trained on 15